diff --git a/docarray/array/storage/base/seqlike.py b/docarray/array/storage/base/seqlike.py
index 5e46cafe607..ce89b82a3bf 100644
--- a/docarray/array/storage/base/seqlike.py
+++ b/docarray/array/storage/base/seqlike.py
@@ -50,7 +50,7 @@ def __eq__(self, other):
         ...
 
     def __len__(self):
-        return len(self._offset2ids)
+        ...
 
     def __iter__(self) -> Iterator['Document']:
         for _id in self._offset2ids:
diff --git a/docarray/array/storage/milvus/seqlike.py b/docarray/array/storage/milvus/seqlike.py
index 1711c5b8080..4ae32860054 100644
--- a/docarray/array/storage/milvus/seqlike.py
+++ b/docarray/array/storage/milvus/seqlike.py
@@ -1,6 +1,6 @@
 from typing import Iterable, Iterator, Union, TYPE_CHECKING
 from docarray.array.storage.base.seqlike import BaseSequenceLikeMixin
-from docarray.array.storage.milvus.backend import _batch_list
+from docarray.array.storage.milvus.backend import _batch_list, _always_true_expr
 from docarray import Document
 
 
@@ -56,3 +56,16 @@ def _extend(self, values: Iterable['Document'], **kwargs):
             payload = self._docs_to_milvus_payload(docs_batch)
             self._collection.insert(payload, **kwargs)
             self._offset2ids.extend([doc.id for doc in docs_batch])
+
+    def __len__(self):
+        if self._list_like:
+            return len(self._offset2ids)
+        else:
+            # Milvus has no native way to get num of entities
+            # so only use it as fallback option
+            with self.loaded_collection():
+                res = self._collection.query(
+                    expr=_always_true_expr('document_id'),
+                    output_fields=['document_id'],
+                )
+                return len(res)
diff --git a/docs/advanced/document-store/extend.md b/docs/advanced/document-store/extend.md
index a65d5ac32bb..591d2ce8832 100644
--- a/docs/advanced/document-store/extend.md
+++ b/docs/advanced/document-store/extend.md
@@ -145,6 +145,9 @@ class SequenceLikeMixin(BaseSequenceLikeMixin):
     def __add__(self, other: Union['Document', Iterable['Document']]):
         ...
 
+    def __len__(self):
+        ...
+
     def insert(self, index: int, value: 'Document'):
         # Optional. By default, this will add a new item and update offset2id
         # if you want to customize this, make sure to handle offset2id
@@ -158,10 +161,6 @@ class SequenceLikeMixin(BaseSequenceLikeMixin):
         # Optional. Override this if you have better implementation than appending one by one
         ...
 
-    def __len__(self):
-        # Optional. By default, this will rely on offset2id to get the length
-        ...
-
     def __iter__(self) -> Iterator['Document']:
         # Optional. By default, this will rely on offset2id to iterate
         ...